R help - Jun 2012 - cointegration, changing time span, incorrect results

I just tried to figure out why R does not calculate the same johansen test
statistics as eviews does:

I imported data:
sy=read.csv("sy.csv",sep=";",header=TRUE)
si=read.csv("si.csv",sep=";",header=TRUE)

merged them.
swe=merge.zoo(si,sy)

when i test swe for cointegration with the johansen test, i get the same
results as in eviews.

however, if i change the time span, both in R and eviews, i get very
different results. i changed the time span in r like this:
first approach:
swe.1=swe[61:216]
second approach:
si.1=(window(si,c(1995,1),c(2007,12)))
sy.1=(window(sy,c(1995,1),c(2007,12)))
swe.2=merge.zoo(si.1,sy1)

if i now test cointegration on swe.1 or on swe.2 (both yield the same
result) I get very different results than eviews suggests.

any comments highly appreciated, i am trying to figure out the difference
for far too long now.

kind regards, Philipp

--
View this message in context:
http://r.789695.n4.nabble.com/cointegration-changing-time-span-incorrect-results-tp4633423.html
Sent from the R help mailing list archive at Nabble.com.

problem solved. if one adjusts in r the time span the way i did it, r really
only has data available for this time span. to the opposite in eviews. if
one sets in eviews a time span via adjusting the sample size, it still uses
the observations outside the sample size i.e. in cointegrationn tests where
lags are used the lags for the first observations are taken from outside the
sample size.

kind regards

--
View this message in context:
http://r.789695.n4.nabble.com/cointegration-changing-time-span-incorrect-results-tp4633423p4633427.html
Sent from the R help mailing list archive at Nabble.com.